AI Roster decisions are awful

rasnell · 07-08-2006, 07:38 AM

I don't know if it's possible, but the patch 2 tweaks that supposedly were going to fix roster management have made it worse in two years of sims and close scrutiny.

The AI is picking horrible starting lineups, starting lower ranked players in every category (offense and defense) where it's not even a close call. It's moving outfielders to out of position spots when both positions had solid stars that should have been in the lineup.

Fatigue and injuries are not factors because I'm talking about opening day lineups.

The draft AI is poor after the fifth round and the human team recommendation button in the draft is broken after five rounds (choosing players that are 1 out of 100 and naming the same one every round thereafter).

My measurement of the toughness of the AI is that I should not always be in first and win the World Series in the very first year of an inaugural draft and I did both times. I'm not that good.

rasnell · 07-08-2006, 08:29 AM

We also have the return of a problem that first started in OOTP 3: The AI is putting the best on-base guy as leadoff hitter -- when this same player is also the BEST POWER hitter and should be in the #4 spot.

I'm sure that's difficult to code and fix, but it eventually got fixed by OOTP 5 and I believe we even had an ability to customize the priority for leadoff hitters according to on-base and speed.

Steve Kuffrey · 07-08-2006, 03:10 PM

Quote:

Originally Posted by rasnell

I don't know if it's possible, but the patch 2 tweaks that supposedly were going to fix roster management have made it worse in two years of sims and close scrutiny.

The AI is picking horrible starting lineups, starting lower ranked players in every category (offense and defense) where it's not even a close call. It's moving outfielders to out of position spots when both positions had solid stars that should have been in the lineup.

Fatigue and injuries are not factors because I'm talking about opening day lineups.

I've just opened up a league created post patch#2 and tried to look at the issues mentioned here in particular and while I did not look at every team I did look at most team's current starting lineups(I think I've gotten to mid-June in the league) and out of the teams I looked at I think that I saw what I thought might have be a questionable lineup decision twice. I looked only at ratings and positions, based on who was on the major league roster. I'm sure that this is an area where constant tweaking and adjusting will take place but I didn't see an over-abundance of issues in the area of starting lineups. I used a standard league setup and just changed the rating systems to 1-100and used scouts and coaches.

rasnell · 07-08-2006, 03:55 PM

My test was based on two full seasons, clean leagues, post patch 2, with close scrutiny of lineups, depth charts, rotation. In addition to very poor choices per position, leadoff hitters were often the best sluggers on the team -- about half of all teams.

battists · 07-10-2006, 03:21 PM

Rasnell,

Based on your previous posts, I'm guessing that your tests were done with a historical league, is this correct? Could you please provide the details of the league structure you are using for these tests, so we can level the playing field?

Let's take one thing at a time. Could you please post an example of a lineup that you thought was really bad, along with the details that show why you thought it was bad? For example, one of these lineups that exhibits the OBP/slugger issue would be good.

Next, could you post an example of this?

Quote:

Originally Posted by rasnell

starting lower ranked players in every category (offense and defense) where it's not even a close call. It's moving outfielders to out of position spots when both positions had solid stars that should have been in the lineup

The draft after round 5 is being discussed in another one of your threads.

Thanks very much!

Steve

rasnell · 07-10-2006, 09:57 PM

Much greater detail posted after a 50-year sim in fictional. Similar problems reported, but far more details. I'll try to find the link to the other thread and post here.

Here's the link:

http://www.ootpdevelopments.com/boar...d.php?t=126634

battists · 07-10-2006, 10:18 PM

Closing this to focus on the other thread Rasnell mentioned.

07-08-2006, 07:38 AM	#1
rasnell Hall Of Famer Join Date: Jan 2003 Location: Frankfort, Kentucky Posts: 3,746	AI Roster decisions are awful I don't know if it's possible, but the patch 2 tweaks that supposedly were going to fix roster management have made it worse in two years of sims and close scrutiny. The AI is picking horrible starting lineups, starting lower ranked players in every category (offense and defense) where it's not even a close call. It's moving outfielders to out of position spots when both positions had solid stars that should have been in the lineup. Fatigue and injuries are not factors because I'm talking about opening day lineups. The draft AI is poor after the fifth round and the human team recommendation button in the draft is broken after five rounds (choosing players that are 1 out of 100 and naming the same one every round thereafter). My measurement of the toughness of the AI is that I should not always be in first and win the World Series in the very first year of an inaugural draft and I did both times. I'm not that good.

07-10-2006, 09:57 PM	#6
rasnell Hall Of Famer Join Date: Jan 2003 Location: Frankfort, Kentucky Posts: 3,746	Much greater detail posted after a 50-year sim in fictional. Similar problems reported, but far more details. I'll try to find the link to the other thread and post here. Here's the link: http://www.ootpdevelopments.com/boar...d.php?t=126634 Last edited by rasnell; 07-10-2006 at 10:03 PM.

07-08-2006, 08:29 AM	#2
rasnell Hall Of Famer Join Date: Jan 2003 Location: Frankfort, Kentucky Posts: 3,746	We also have the return of a problem that first started in OOTP 3: The AI is putting the best on-base guy as leadoff hitter -- when this same player is also the BEST POWER hitter and should be in the #4 spot. I'm sure that's difficult to code and fix, but it eventually got fixed by OOTP 5 and I believe we even had an ability to customize the priority for leadoff hitters according to on-base and speed.

07-08-2006, 03:55 PM	#4
rasnell Hall Of Famer Join Date: Jan 2003 Location: Frankfort, Kentucky Posts: 3,746	My test was based on two full seasons, clean leagues, post patch 2, with close scrutiny of lineups, depth charts, rotation. In addition to very poor choices per position, leadoff hitters were often the best sluggers on the team -- about half of all teams.

07-10-2006, 10:18 PM	#7
battists Hall Of Famer Join Date: Jul 2004 Posts: 18,506 Blog Entries: 7	Closing this to focus on the other thread Rasnell mentioned.